Steve Yegge, Sourcegraph, ETLS Vegas (21)
summary
Claims that LLM-based coding assistants and an iterative “chop” workflow make developers 5–10x faster by generating code and then digging back into the codebase; IDEs are limiting compared with LLMs and require new pipelines to process IP. Organizations face thousands of repos and content in wikis/issues, with one example ingesting 50 pyes a day, and most teams are not set up to discover or feed that scale into LLMs yet.
AI-written summary · about
related talks
54mTech Truth: Agile Evolution & the Future of SW Engineering
39mAI Assistance Beyond Code: What Do We Need to Make it Work?
32mFrom Vibes to Outages: Riding the AI Code Wave
20mFireside Chat
43mAI & HR
25mHywel Carver at LeadDev West Coast 2023
01h 01mYour code is just a detail
22mThe AI Productivity Paradox in Engineering